Genetics of common complex psychiatric disorders II

Structure of DNA

Gene structure

Types and sizes of genetic variants

Chromosomes

Diploid (two copies of each chromosome)

Single Nucleotide Polymorphisms (SNP)

Single base of DNA is substituted with another.
Occur in 1 of every 100 - 1000 basepairs (bp).
Currently, ± 1.2B SNPs are known.
Labelled with a dbSNP Reference ID (“rs number”)

Canidate gene studies (circa 2000)

Example: Catechol-O-methyltransferase (COMT)

SNP rs4680 G>A substitution (missense variant) results in amino acid change from valine to methionine at codon 158 (Val158Met).

COMT is involved in inactivation of dopamine, epinephrine, and norepinephrine. The Val variant metabolises dopamine at a higher rate than the Met variant. It was proposed that the rs4680 SNP was related to a large number of psychiatric disorders, but these results have not replicated.

Another highly studies variant was 5-HTTLP (serotonin-transporter-linked polymorphic region) in SLC6A4 (serotonin transporter gene).

“linkage studies investigate correlations between a disease and inheritance of specific chromosomal regions in families, whereas association studies focus on differences in the frequency of specific genetic markers in groups of affected and unaffected individuals.”

Sullivan … Neale (2001) Genetic Case-Control Association Studies in Neuropsychiatry, ARCH GEN PSYCHIATRY 58
Border et al. No Support for Historical Candidate Gene or Candidate Gene-by-Interaction Hypotheses for Major Depression Across Multiple Large Samples. American Journal of Psychiatry doi:10.1176/appi.ajp.2018.18070881

Genome-wide associations studies (GWAS) (2007 - today)

Candidate gene findings did not replicate
Power analysis: candidate gene study sample sizes were vastly too small
Alternative: Assay the whole genome (hundreds of thousands or millions of variants)
Hypothesis-free testing

Conducting a GWAS

Two alleles: A and a
Three genotypes: AA, Aa, aa
Arbitrarily select one allele (A) as the effect allele
For each subject, count number of effect alleles: \(x = 0, 1, 2\)

Regress on phenotype \(Y\)
Linear regression: \(Y \sim \alpha + \beta x + \mathrm{covariates}\)
Logistic regression: \(\mathrm{Prob}(Y= 1) \sim \mathrm{logit}^{-1}(\alpha + \beta x + \mathrm{covariates})\)
Repeat for all markers

To run the regression, the SNP genotype is encoded to a number as a count of one of the alleles (which allele is chosen is arbitrary, but traditionally it is allele with the lower frequency, called the minor allele). The chosen allele is called the “effect allele”. In this example, if the T allele is chosen as the effect allele, then the numeric encoding is the number of T alleles in the genotype: CC = 0, CT and TC = 1, and TT = 2. This numeric representation can then be entered into a regression formula.

From the fitted regression model, the beta coefficient \(\beta\) for the SNP represents the average substitution effect of having one additional copy of the effect allele.

Binary traits like disease status are analysed as case-control studies using logistic regression. In a case-control analysis, the beta coefficient represents the change in log-odds of having the disease for every additional copy of the effect allele. Exponentiating the beta coefficient, \(\exp(\beta)\), converts the effect size to an odds ratio.

Balding, D. A tutorial on statistical methods for population association studies. Nat Rev Genet 7, 781–791 (2006). doi:10.1038/nrg1916
Uffelmann, E., et al. Genome-wide association studies. Nat Rev Methods Primers 1, 59 (2021). doi:10.1038/s43586-021-00056-9

15+ years of GWAS

GWAS data

CHROM	POS	ID	REF	ALT	P	OR	AF
chr1	1006415	rs145588482	TGGCAGCTC	T	0.0004688	0.583384	0.01293
chr1	1007256	rs76233940	G	A	0.0004016	0.579783	0.01301
chr1	1341559	rs376494450	C	T	0.0001512	1.320130	0.02706
chr1	1480224	rs71628972	A	G	0.0007602	1.412750	0.01319
chr1	2035619	rs111334586	C	T	0.0009206	1.413370	0.01209
chr1	2213349	rs262695	A	G	0.0008815	0.904873	0.32073